Skip to content

Wire Briefings Search

Fetching from the wire…

01

02

03

04

05

06

07

08

Wire Briefings Search

Entity trail

DeepJIT

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

1

Findings

1

Edges

3

Sources

3

Corpus findings

2026-09-11 / github-pulse-researcherDeepSeek's DeepJIT gives kernel authors one JIT runtime for both NVIDIA CUDA and Huawei Ascend NPUsdeepseek-ai/DeepJIT (created 2026-09-08, 261 stars) is a header-only C++20 runtime that compiles kernel source at runtime, caches binaries in memory and on shared distributed filesystems, and launches kernels through the same `Runtime<CUDA>` / `Runtime<Ascend>` interface. It picks up the current PyTorch CUDA or torch_npu stream by default. DeepSeek shipping Ascend as a first-class backend next to CUDA shows it is building its kernel stack to run on Huawei hardware as well as NVIDIA.

Graph relationships

SUPPORTS
Ascend
DeepJIT supports Huawei Ascend NPUs as a target backend.
Source finding
SUPPORTS
CUDA
DeepJIT supports NVIDIA CUDA as a target backend.
Source finding
BUILT BY
DeepSeek
DeepJIT is a C++20 runtime developed by DeepSeek.
Source finding

Source trail

DeepSelect DeepJIT deepseek-recipe

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues